Fundamental Limits of Ultra-Reliable Low-Latency Communication

Doctoral thesis, 2024

This thesis focuses on the fundamental performance of ultra-reliable lowlatency
communication (URLLC) systems, particularly in terms of achievable
error probabilities. Since low latency necessitates the use of short data packets,
understanding the trade-off between blocklength and reliability is crucial.
This thesis provides design guidelines for URLLC systems, including those
with multiple antennas, imperfect time synchronization, and error detection,
based on three key papers.
In Paper A, we present a numerically efficient method for evaluating the random
coding union bounds with parameter s for the error probability achievable
by a pilot-assisted transmission method on a block-fading channel. Our approach,
which uses the saddlepoint approximation with respect to the number
of fading blocks, significantly reduces the number of Monte Carlo samples required
to accurately estimate the achievability bound for the error probability,
particularly in scenarios with multiple diversity branches.
In Paper B, we investigate the achievable error probability in the finite
blocklength regime for a pilot-assisted transmission scheme operating over an
imperfectly synchronized, memoryless block-fading waveform channel. Numerical
experiments show that the number of pilot symbols needed to estimate
the fading channel gains with the accuracy required for URLLC is also sufficient
for synchronization when delays are modeled as fully dependent across
fading blocks.
In Paper C, we explore a URLLC system with an erasure decoder and
investigate the trade-off between the total error probability and the probability
of undetected errors in the short blocklength regime by developing two
achievability bounds. These bounds are tighter than Forney’s error exponent
bound for the additive white Gaussian noise channel and can be evaluated for
practically relevant channels.